An FM IF Filter Wobbulator

For another project (FM Tuner Project, still to be finalized) I needed to be able to align the IF filter. Now I don't have a sweep set like the Rohde & Schwarz SWP/ZAS combination (it's #1 on my want list!) so I decided to make my own. I got my inspiration from an ancient (1980) German publication called UHF Unterlage. Choc full of designs for the UHF amateur it also includes a design for a wobbulator, also known as a sweep generator. I also happen to have a couple of AD8307ARs, the famous Analog Devices logarhythmic demodulator, all but screaming to be used in an application like this.

So what does it do? The wobbulator generates a swept RF signal which is sent through the DUT (Device Under Test, usually a filter or otherwise frequency dependent device) and then demodulated and displayed on a screen. A scope already has a sawtooth oscillator together with associated blanking circuitry. However, my trusty Philips PM3055 doesn't have a sweep output (ot's optional, though). So I kludged together a video amp with an LT1252 and was done.

A blown up server power supply would be a nice case, providing some shielding and having a power entry already available. The power supply had to be complicated. Because I wanted double voltages, 15 and 30 V, I could not use two 7815s or the ubiquituous 7815/7915 combo, so I chose a 7815 which also is the reference for a second stabilizer with a TIP31C. Works like a charm.

To make my life even harder I decided to use the video amplifier just as one and have the sweep input terminate into 75 ohms. But since the JFET TL072 opamp doesn't sense near the negative power rail I needed to lift the sweep. This produced some headaches but eventually it was solved.

The VCO is not linear so I needed an predistortion circuit. Using Linear Technology's LTSpice I designed a linearizer. In the computer it worked beautifully but in real life the performance is a bit lacking. But as I'll use a marker generator I decided to leave it. The VCO itself caused more headaches and I finally decided to for an "ECO" (Electron Coupled Oscillator) type circuit with a tap on the oscillator inductor. This is a 45 turns coil on a 5 mm bobbin without core. Taking the signal from the drain was not an option as harmonic content was huge. Taking it from the tank itself with a second JFET as buffer worked much better. Even so, a fifth order Chebyshev filter was needed to produce a clean signal.

In UHF Unterlage the marker mixer is a diode but I wanted more isolation so I chose a BF981 MOSFET. The mixer produces an LF beat which is processed by the marker amp. The second half of the TL072 amplifies the beat product which is voltage doubled and rectified and sent to a precision timer (vintage HEF4538BP). Two timers are in series to prevent multiple markers. The second timer is 0.2 ms which is used to blank the CRT in the scope. Initially the idea was to superimpose the marker onto the demodulated signal but this didn't work too well. Now the trace is blanked during the marker. This works well.

Finally the demodulator. This is dedicated for the high-impedance filter that I want to build, and uses a JFET buffer before the AD8307. The marker input is not connected, and the 5 V is regulated with a TL431 shunt regulator because the connection is made with a 3.5 mm jack which could cause glitches.

After a weekend screaming and cursing my wobbulator was done. Time to test it out with an old SFE10.7MA ceramic filter. Immediately it was clear that the VCO didn't produce enough juice. I toyed with the idea of building a 20 dB amp inside the box but then I remembered my CATV wideband amplifier module with a BGY55. While not characterized for use below 47 MHz it is flat to below 10 MHz and indeed it works like a charm. I can now measure about 50 dB down. The leakage from the marker input is enough to lift the noise floor a few dBs, so it must be switched off.

With the scope adjusted to 0.25 V/div it's linear with 10 dB/div which is sufficient for this purpose. But once I have a proper sweeper it will be retired to the junk box for sure!


sch_sweep_buffer

sweep_buffer

sweep_output_bnc

sweep_out

sch_psu

sch_vco

sch_vco_linearizer

sch_marker_amp

sch_probe

psu

signal_processor

vco_final

vco_filtered

vco_marker_mixer

finished_boxie

connections

dut

probe

marker_source

baseline_marker

dut_10.5_marker

dut_10.7_marker

dut_10.9_marker

dut_6dbm

back

19 October 2015